Vendors Deliver On The Promise Of Self-Managing AppsVendors Deliver On The Promise Of Self-Managing Apps

For a few years, vendors have been promising software that can troubleshoot and automatically fix systems, then apply the lessons learned to the next problem that arises. These self-managing systems are still works in progress, but they're getting closer to making good on those promises.

BMC Software Inc. and Computer Associates offer products that let companies proactively manage their IT infrastructures. Others, including Hewlett-Packard, IBM, Mercury Interactive, Network Physics, and Rainfinity, are also working with the technology, analysts say. The goal? To recognize and fix a problem before it affects business.

This technology is vital to the success of Internet Operations Center Inc., an E-business service provider in Southfield, Mich. "That predictive capability is unequivocally part of our infrastructure today," CFO Dave Marczak says. With more than 1,500 customers and only about 70 IT specialists, the company can use a hand monitoring and managing the Internet services it provides.

Internet Operations Center turned to BMC, whose Perform software helped it sort out a performance problem on a network of six servers that support 2,500 concurrent users. Perform had been tracking processes running on the servers' hard drives, memory boards, and CPUs. When performance started to degrade, Perform was able to determine how each process was behaving and correlate that with statistical data that the software had collected during previous load testing. "We saw that processes never consumed more than 50% of server capacity," says Dave Castano, an application deployment specialist at Internet Operations. "Perform figured out that it was a network problem, not a server problem, within a couple of hours." Castano estimates that Perform saved his company 200 hours of work.

The BMC products at Internet Operations include Patrol Perform, which provides advanced performance-analysis capabilities for optimizing a server environment, and Patrol Predict, which delivers advanced prediction capabilities to plan for changes. Together, the BMC products help Internet Operations' customers understand the past and present performance of their server environments. The built-in analysis and detailed representation of the server environment is helpful to Internet Operations. "BMC already created the knowledge base in Perform," Castano says.

David Wagner, BMC's director of product marketing, says the knowledge modules inside Patrol Predict helps customers plan for disaster recovery. The BMC software kicks off load tests that simulate what happens when a server goes down, and that information is used to build necessary redundancies. The software also lets IT managers simulate network and system spikes to determine thresholds, which, when reached, triggers automatic actions, such as switching work to other servers.

CA first began pitching its self-managing and predicative-analysis software, called Neugents, in 1999. CA originally had offered Neugents for its Unicenter enterprise management system, but the technology has also appeared in CA's security, data-management and application development, storage, portal, business-intelligence, and application life-cycle-management product lines.

Despite several years' development, Neugents are still being fine-tuned. They're being used to tell IT managers why a router is the bottleneck on a stalled network by comparing how the router is running with data compiled when the network was operating optimally, says Carl Hartman, CA's VP of brand management. CA is testing Neugents that can identify a virus before it gets into a system and ones that can determine when a network service deviates from a service-level agreement, he says.

Simon Samaha, VP of IT and CIO at $420 million Cooper Health System in Camden, N.J., likes what CA's predictive technology does for his IT infrastructure. Now he hopes to apply the same technology to business processes. Samaha wants to use Neugents to increase synergy among hospital departments, revenue, and patient care. To do that, the health-care company is building data models using Neugents that will predict outcomes to certain scenarios. One such model helps determine the departments and costs that will be involved and the treatment a patient will require by comparing that patient's symptoms against the data.

"So far, one of the best achievements I've seen is Neugents' ability to cut data-preparation time from weeks down to hours," Samaha says. Still, the project will take time: Samaha hopes for results in the next six to eight months.

Self-managing and predictive-analysis technology is improving, says Richard Ptak, a Hurwitz Group analyst. IT managers have a vast amount of management data, and self-management tools are increasingly able to process and act on that information, he says.

The improvements, though slow in coming, are worth the wait. Says Michael Dortch, a Robert Frances Group analyst, "There could be mice on treadmills inside the servers, but as long as those servers deliver useful, actionable information that helps improve business competitiveness or efficiency, they will be embraced by enterprise IT executives."